Search

​

This SAFER-SIM project will use simulation technology to provide a framework for testing and evaluating autonomous vehicles in a virtual environment. The focus of the framework is to assess vehicle behavior in cost prohibitive or safety critical scenarios, e.g. an animal running onto the road during inclement weather where the physics or sensing may limit the capabilities of the vehicle. A critical component of this autonomous vehicle simulation framework is the ability to synthetically generate sensor data for software-in-the-loop testing. This sensor data should be physically realistic such that true limitations can be explored. For example, a scenario involving rain, snow, or fog may result in limited sight from a camera. For understanding what happens if it unexpectedly begins to rain, the virtual camera sensor should mimic the distortion from the rain on the camera, or the limited range of sight through the raindrops. This project’s objective is to develop and implement physics-based camera and LiDAR sensor models, validate them against data captured by physical sensors, and demonstrate their use within the Chrono CAV platform on realistic mixed autonomous/conventional traffic scenarios.